In vitro electrophysiology

Mice aged between P50 and P60 were killed by cervical dislocation, decapitated and the brain quickly removed and cooled in ice-cold dissection buffer (in mM: 108 choline-Cl, 3 KCl, 26 NaHCO₃, 1.25 NaH₂, PO₄, 25 D-glucose, 3 Na-pyruvate, 1 CaCl₂, 6 MgSO₄, 285 mOsm) bubbled with 95% O₂/5% CO₂. Coronal slices (350 µm thick) were cut on a vibrating microtome (Microm HM650V, Thermo Fisher, Cheshire, UK) and transferred to a holding chamber containing normal artificial CSF (in mM: 119 NaCl, 3.5 KCl, 1 NaH₂PO₄, 10 D-glucose, 2 CaCl₂, 1 MgSO₄, 300 mOsm) bubbled with 95% O₂/5% CO₂. Slices were incubated at 32°C for 45 min after slicing, then returned to room temperature until recording.

Whole cell recordings were performed at 35–37°C. Barrels were identified in slices under bright field illumination using an Olympus BX50WI microscope. Pyramidal neurons were identified using DIC optics. Recording pipettes (4–10 MΩ) were pulled from borosilicate glass (Clark GC150-F10, Harvard Apparatus, UK) and filled with a potassium gluconate-based recording solution (in mM: 110 K-gluconate, 10 KCl, 2 MgCl₂, 2 Na₂ATP, 0.03 Na₂GTP, 10 HEPES, pH 7.3, 270 mOsm). LII/III pyramidal cells were selected for their characteristic regular spiking behavior under depolarizing current. Recordings were aborted if Vm deviated spontaneously by more than 5 mV, or access resistance deviated by more than 20% during the recording. An Axon Multiclamp 700 B (Molecular Devices, Sunnyvale, CA) in current clamp mode was used as the patch amplifier and signals were telegraphed to and digitised by a CED 1401 with Signal software (CED, Cambridge, UK) running on a Windows PC.

A tungsten monopolar stimulating electrode was placed centrally to a selected barrel in layer IV and suitable pyramidal neurons were identified in the area vertically above the stimulating electrode. Once a whole cell recording was established extracellular stimulus was applied at 0.1 Hz, consisting of a pairs of pulses at 20 Hz. The stimulus was of 0.2 ms duration and 1–35 V intensity, designed to produce a monosynaptic EPSP of 3–6 mV in the postsynaptic cell. After 10 min of baseline recording, LTP was induced by pairing a suprathreshold 2.5 ms somatic depolarizing pulse with a single presynaptic stimulus (5 ms pre-post interval). Four runs of 50 paired stimuli at 2 Hz were delivered at 0.025 Hz. After LTP induction the stimulus paradigm was switched back the same as in the baseline recording and one hour of post-paring data were acquired. EPSP amplitudes were calculated as the peak response above the preceding baseline voltage. Significant potentiation was calculated by comparing mean EPSP amplitude at 50–60 mins after LTP induction with mean baseline EPSP amplitude using Student’s t-test.

The position of the recording and stimulating electrodes were identical to those for the LTP experiments. Once the recordings were established the perfusion solution was switched to magnesium-free ACSF and an NMDA-receptor mediated response was isolated by the addition of 20 µM CNQX to the bath solution. Single extracellular stimuli were delivered at 0.1 Hz. After a stable baseline had been established, stimulation was halted and 10 µM MK-801 was washed onto the slice for 10 min. Stimulation was then resumed as before and at least 100 further stimulus trials were recorded, still in the presence of MK-801 and CNQX in Mg-free ACSF. Single and double-exponential fits of the response amplitude over time were performed with GraphPad Prism 5 and 6 (GraphPad, La Jolla, CA). The rate of decrease in the NMDAr-mediated EPSPs is directly related to the release probability of synapses in the observed pathway.

Recordings were made in layer II/III pyramidal cells. Miniature AMPA receptor-mediated EPSPs were isolated by the bath application of 1 µM tetrodotoxin, 100 µM picrotoxin and 50 µM D-AP5. Action potential blockade was confirmed with the injection of highly-depolarizing square current pulses (0.8–1 nA, 500 ms). 100–500 events were analysed per cell with a template-matching and threshold-crossing method (Axograph X, Berkeley, CA). Well-defined, isolated mEPSPs were identified and used to train a loose-fitting template, with a detection minimum threshold of 2.5 times the RMS baseline noise (Clements and Bekkers, 1997). Amplitudes and inter-event intervals were binned and their cumulative distribution compared using a Kolmogorov-Smirnov test.